The empirical volatility-growth relationship: is economic freedom the missing link?

Author:Dawson, John W.
  1. Introduction

    Ramey and Ramey (1995) is widely regarded as the benchmark empirical study of the relationship between business cycle volatility and long-run economic growth. The Rameys report a negative relationship between volatility and growth in a broad cross-section of countries. A number of more recent studies confirm the finding of a negative volatility-growth relationship, including Martin and Rogers (2000), Fatas and Mihov (2003), Mobarak (2005), Badinger (2010), and Furceri (2010). Still other recent studies, however, report a positive relationship; see, for example, Stastny and Zagler (2007) and Lee (2010). Two studies that predate Ramey and Ramey--often ignored in this literature--also examine the volatility-growth relationship. In their broad search for determinants of cross-country I growth, Kormendi and Meguire (1985) and Grier and Tullock (1989) find evidence of a positive relationship between volatility and growth.

    At first glance, it seems that the available evidence on the empirical relationship between volatility and growth is contradictory. A closer look at the evidence, however, reveals some interesting patterns in the data. First, Ramey and Ramey's finding of a negative relationship in a ninety-two-country sample becomes positive and statistically insignificant when their sample is restricted to countries belonging to the Organisation for Economic Co-operation and Development (OECD). Similarly, Kormendi and Meguire's positive relationship is obtained in a sample of forty-seven mostly developed countries, and the positive relationships found in the studies by Stastny and Zagler and by Lee are obtained in samples restricted to OECD and Group of 7 (G7) countries. In addition, Dawson and Stephenson (1997) find no evidence of a volatility-growth relationship across the U.S. states. All of this information suggests that the generally accepted negative relationship between volatility and growth may not be an accurate description of the process at work in more developed economies.

    It is interesting to consider which characteristic of more developed economies drives this pattern. Specifically, this paper considers whether the volatility-growth relationship varies with levels of economic freedom across countries and whether volatility is serving as a proxy for economic freedom in studies of the volatility-growth relationship. It is well known that economic freedom is an important determinant of growth across countries; see, for example, studies by Dawson (1998) and by Gwartney, Lawson, and Holcombe (1999), among many others. (1) More recently, Lipford (2007) and Dawson (2010) have shown that economic freedom is also related to business cycle volatility across countries. It is possible, then, that volatility is serving as a proxy for economic freedom in studies of the volatility-growth relationship that do not explicitly control for differences in freedom across countries. It is also possible that volatility and growth are positively related or unrelated in countries with higher levels of economic freedom and negatively related in countries with lower levels of freedom. Such possibilities could explain why volatility and growth are negatively related in diverse samples of countries, but found to be positively related or 1 insignificant in samples restricted to more developed countries where freedom is at a higher and more uniform level. It can also explain why volatility and growth are not related across the U.S. states where freedom is also at a higher and more uniform level.

    Evidence provided by Grier and Tullock also supports this conjecture, where a positive volatility-growth relationship is found in a large, diverse sample of countries using a specification that includes several institutional proxies--and the size and significance of the volatility coefficient is reduced when an explicit measure of institutions is included. Studies that find a negative relationship between volatility and growth in broad samples of countries may also be consistent with the idea that economic freedom matters in the volatility-growth relationship. If such studies ignore the role of freedom, the analysis may attribute to volatility the influence that is actually due to freedom.

    This paper explores the possibility that economic freedom is the missing link in the relationship between macroeconomic volatility and economic growth. The next section provides a brief theoretical perspective on the volatility-growth relationship. The third section discusses the empirical model, methodology, and data in detail. A discussion of the empirical results appears in the following two sections, and the final section offers concluding remarks.

  2. Volatility and Growth: A Brief Theoretical Perspective

    In terms of theory, relatively little attention has been paid to the effect of business-cycle volatility on long-run economic growth. Indeed, the literatures on business cycles and economic growth have existed largely in isolation from one another. There are reasons, however, to believe that volatility and growth may be related. For example, economic uncertainty and credit constraints during periods of increased macroeconomic volatility may reduce investment, capital accumulation, and presumably growth. Along similar lines, if investment is to some extent irreversible, increased volatility can lead to lower investment and thus lower growth; see, for example, Bernanke (1983). Both of these channels suggest a negative relationship between volatility and growth.

    There are also reasons to suspect a positive relationship between volatility and growth. Black (1987) suggests that economies face a positive risk-return trade-off where riskier technologies (that ultimately lead to higher volatility) are adopted only if they are expected to pay a higher return and hence produce higher growth rates. Separately, Sandmo (1970) and Mirman (1971) hypothesize that more variable income streams lead to higher savings, more investment, and presumably more growth. These channels both imply a positive volatility-growth relationship.

    Clearly, there are different possible channels through which volatility may affect growth, some with positive and some with negative predicted relationships. In addition, different channels may be dominant in different economies, causing different estimated relationships in different groups of countries. Which channel dominates in an economy may well depend on certain characteristics in that economy. In particular, different institutional arrangements may determine which channel is dominant. For instance, economies with more market-oriented institutions (i.e., more economic freedom) may be able to adjust to volatility more readily, thus mitigating the negative effect of volatility on investment. This arrangement may, in turn, result in a statistically insignificant or positive estimated volatility-growth relationship in high-freedom countries. Similarly, myopic behavior in countries with low levels of economic freedom may dampen precautionary saving motives even in times of high volatility, thus reducing the positive influence of volatility on growth. This structure could leave a negative volatility-growth relationship at work in these countries.

    While the preceding theoretical discussion is obviously far from complete, the point is to illustrate that theory alone cannot settle the debate over the relationship between volatility and growth. Moreover, the question of which theoretical relationship emerges in an economy may depend on the institutional framework. Ultimately, it is an empirical issue. The analysis in the remainder of the paper addresses this empirical question.

  3. Empirical Model, Methodology, and Data

    The following empirical specification is typical of that used in studies of the volatility-growth relationship:

    [DELTA]ln [y.sub.i] = [alpha] + [lambda][[sigma].sub.i] + [[summation].sub.j][[beta].sub.j][X.sub.ji] + [[epsilon].sub.i].

    The dependent variable, [DELTA]ln y, is the average annual growth rate of real GDP per capita. [X.sub.j] represents a common set of conditioning variables found by Levine and Renelt (1992) to be robustly related to growth. These conditioning variables include the initial income level, the investment share of GDP, and population growth. [sigma] is the volatility measure and [lambda] is the coefficient of interest. This basic specification is used as a starting point in the analysis that follows.

    The explanatory variable of interest, macroeconomic volatility, is measured using the standard deviation of annual growth rates of real GDP per capita. This is a standard measure of business cycle volatility that has been used in a number of recent studies, including the pure cross-section specification in Ramey and Ramey. This volatility measure implicitly assumes that the trend growth rate is constant and equal to the mean for each country. (2)

    To determine whether the volatility-growth relationship varies across countries with different institutional environments, measures of economic freedom are added as explanatory variables in the specification above. In regressions that include economic freedom, both the initial level of freedom and the change in freedom over the sample period are included. Changes in economic freedom have been shown to be important, along with the level of freedom, in explaining long-run growth experiences across countries in a number of studies (see, e.g., Dawson [1998]).

    In addition, Pitlik (2002) shows that a measure of the volatility of economic freedom over time is negatively related to long-run growth rates across countries even after controlling for other factors related to growth, including the level and changes in freedom. This result shows that volatile liberalization policies depress growth even when they generally tend toward increased levels of economic freedom. It seems particularly important to control for volatility in the path toward freedom...

To continue reading